A study funded by CFIR asked 68 Canadian startups building on artificial intelligence what it changed for them, and heard more about how the work gets done than about who buys it.
Four in five reported their own work improving: data handled faster, forecasts sharper, costs lower. Roughly two in three reported a larger customer base. The interviews put those in the same order — the savings inside a company arrive first, and customers follow later, within the period the study covers.
The research comes from the Entrepreneurship Research Centre at YEDI, an arm's-length unit of the accelerator. The companies work in healthcare, finance, transportation and agribusiness, and fifteen of their founders and CTOs were interviewed in the Toronto–Waterloo and Vancouver clusters.
Alongside the survey, national figures from 2010 to 2024 show research spending and operating profit rising together in Canada's ICT sector, closely enough that a simple model tracks most of the movement. The paper is careful with this: it calls the pattern a relationship rather than a demonstrated cause, and names what sits outside the model — company size, stage, sector dynamics, region.
It is equally plain about what adoption brings with it. Work disappears as processes automate, decisions get made by systems that are hard to explain, and the distance grows between companies that can reach the best tools and companies that reach for what they can afford.
The paper is published open access. Read it in full, with the figures, the method and the licence.
